Fake Elon Musk, fake Tesla flying car demo & a crypto scam: How Coinbase stopped a $4.7 million fraud (but couldn't save all)
A recent scam uncovered by Coinbase involved a deepfake impersonation of Elon Musk, luring viewers with promises of doubling their cryptocurrency investments through a fake Tesla giveaway livestream. With around 500,000 viewers engaging, many fell...

The scam involved an AI-generated deepfake of Tesla CEO Elon Musk during a purported Tesla flying car demonstration.
The scam involved an AI-generated deepfake of Tesla CEO Elon Musk during a purported Tesla flying car demonstration. The fake livestream appeared similar to an authentic broadcast and displayed cryptocurrency wallet addresses and a QR code. Viewers were promised that their cryptocurrency would be doubled if they sent funds to the displayed wallet addresses. The event was watched by more than 500,000 viewers, prompting hundreds of Coinbase customers to attempt cryptocurrency transfers worth millions of dollars. Coinbase said hundreds of its customers attempted to send money to the scammers within approximately two hours of the livestream.
How Coinbase's AI system stopped $4.7 million in crypto transfers
Coinbase said its conventional fraud prevention systems typically focus on detecting transactions that customers have not authorised.However, in this case no Coinbase accounts were not compromised, nor their passwords were stolen, or they were tricked into revealing their login credentials. Instead, customers voluntarily initiated cryptocurrency transfers because they believed the giveaway was genuine.
To address this, Coinbase deployed an AI-powered scam prevention agent that intervened when customers attempted suspicious transactions.
The system used three main components:
1. Real-time scam detection
Machine learning models analysed cryptocurrency transfers in real time to identify suspicious patterns.
These included giveaway-related transactions, brand impersonation attempts and transfers involving newly created wallet addresses.
2. AI-powered customer intervention
When the system flagged a suspicious transaction, an AI agent initiated a targeted conversation with the customer.
It explained the scam pattern, highlighted how cryptocurrency giveaway fraud works and asked questions designed to help customers recognise the warning signs before confirming the transfer.
3. Confidence scoring
Coinbase also assigned a confidence score to every response generated by the AI agent.
The danger of deepfakes
Coinbase said deepfakes make convincing scams cheap, fast, and easy to relaunch under a new name the moment one is shut down."Agents had never seen this scam. Deepfake giveaways tied to a live product launch weren't in its training. It recognized the shape of the scheme anyway, a promise to multiply funds, the urgency of an event that would end in hours, a transfer that could not be reversed, and it ran from the first flagged send to the last, with no humans in the loop," Coinbase added.
To prepare its defenses against such scams, Coinbase said it is now investing heavily in AI to create models that get better through AI-driven research and review suspicious transactions across different scam categories and using AI to label new fraud cases.
Coinbase also urged users not to send crypto to an address or QR code shown in a video, livestream, or message.
"Nobody legitimate will double your crypto. An offer that requires you to send funds first to receive more back is a scam, no matter who appears to be making it," it added.
The Economic Times Business News App for the Latest News in Business, Sensex, Stock Market Updates & More.
The Economic Times News App for Quarterly Results, Latest News in ITR, Business, Share Market, Live Sensex News & More.